Transcript 下載/瀏覽
Robot and Servo Drive Lab. Sensorless IPMSM Drive System Using Saliency Back-EMF-Based Intelligent Torque Observer With MTPA Control IEEE TRANSACTIONS ON INDUSTRIAL INFORMATICS, VOL. 10, NO. 2, MAY 2014 Faa-Jeng Lin, Senior Member, IEEE, Ying-Chih Hung, Student Member, IEEE, Jia-Ming Chen, and Chao-Ming Yeh Advisor: Prof. Ming-Syhan Wang Student: Ika Noer Syamsiana 2016/7/14 Department of Electrical Engineering Southern Taiwan University of Science and Technology Outline : Introduction Saliency Back-emf-based Rotor Fluxangle And Speed Estimator With MTPA Control WFNN Torque Observer Design And Experimentation Conclusion 2016/7/14 2 Department of Electrical Engineering Southern Taiwan University Robot and Servo Drive Lab. INTRODUCTION The permanent magnet synchronous motors (PMSMs) can basically be divided into three categories: 1) surface mounted; 2) inset; and 3) interior Moreover, due to the interior PMSM (IPMSM) has many attractive characteristics such as high-power density, hightorque-to-inertia ratio, wide speed operation range, and free from maintenance. Department of Electrical Engineering Southern Taiwan University INTRODUCTION IPMSM inverter-fed compressor drive system However, the compressor is usually operated in a high-temperature environment with corrosive refrigerant. The hall sensors and encoder cannot be installed in the compressor Department of Electrical Engineering Southern Taiwan University ARCHITECTURE OF SENSORLESS MTPA FIELDORIENTED CONTROL SYSTEM FOR IPMSM Department of Electrical Engineering Southern Taiwan University Estimation of Rotor Flux Angle and Speed The mechanical model-based PLL can be divided into two parts: 1) the torque observer; and 2) the mechanical model of the IPMSM. Department of Electrical Engineering Southern Taiwan University Estimation of Rotor Flux Angle and Speed IPMSM in synchronous rotating reference frame v d rs pLd v L q re d re Lq id 0 rs pLq iq re pm Can be expressed in the stationary reference frame re Lq Ld i v rs pLd sin re i re pm Lq Ld reid piq v L L r pL cos re d q s d re E sin re E re pm Lq Ld re id piq cos re E sin re E E cos re Department of Electrical Engineering Southern Taiwan University Estimation of Rotor Flux Angle and Speed E re pm Lq Ld reid piq can be estimated using a state filter based on current observers in stationary reference frame The saliency back-EMF state filter consists of two parts: 1. the IPMSM model without saliency back-EMF term and 2. two PI current compensators Department of Electrical Engineering Southern Taiwan University Estimation of Rotor Flux Angle and Speed The estimated saliency back-EMF will approximate the actual saliency back-EMF of the IPMSM Eˆ t sin re t lim E ˆ t E t cos re t Eˆ # sin E re ˆ ˆ # sign( re ) E E E cos re # # ˆ ˆ ˆ E cos re E cos ˆre Eˆ sin re ˆre Department of Electrical Engineering Southern Taiwan University Estimation of Rotor Flux Angle and Speed The rotor speed and angle estimation transfer function is given as follows : ˆ rm ˆrm rm rm Ld Lq 3 J s Kd s 2 K p s Ki Ld Lq Department of Electrical Engineering Southern Taiwan University Js3 Kd s 2 K p s Ki Saliency Back-EMF-Based MTPA Control (a) IPMSM MTPA control trajectory Department of Electrical Engineering Southern Taiwan University Saliency Back-EMF-Based MTPA Control (b) Saliency back-EMF-based MTPA control. Department of Electrical Engineering Southern Taiwan University Saliency Back-EMF-Based MTPA Control The torque equation in the d q axis synchronously rotating reference frame of the IPMSM is given as follows : Tem 3P pm iq Ld Lq id iq , excitation torque 4 Tem 3P 3P Ld Lq id iq , reluctance torque pm iq 4 4 The stator phase current can be obtained as follows : I s id2 iq2 Department of Electrical Engineering Southern Taiwan University Saliency Back-EMF-Based MTPA Control Assuming the absolute value of the stator current is keeping constant below its maximum value in the constant torque region of the IPMSM 2 q i dTem pm Ld Lq id Ld Lq 0 diq id id pm Department of Electrical Engineering Southern Taiwan University 2 pm 4Ld L 2Lq Ld i 2 2 q q Saliency Back-EMF-Based MTPA Control The estimated rotor flux angle equals to the real rotor flux angle, the saliency back-EMF use MTPA control using saliency back-EMF state filter Eˆ Eˆ sin ˆre Eˆ cos ˆre 2 2 ˆ E E sin re E cos re E Department of Electrical Engineering Southern Taiwan University Saliency Back-EMF-Based MTPA Control i 2 d L pm d Lq id i 0 2 q 2 Eˆ re pm piq Lq Ld Ld Lq 2 id i q pm re Where Eˆ re pm Ld Lq piq Department of Electrical Engineering Southern Taiwan University The proposed new MTPA control is the saliency back-EMF correction term WFNN TORQUE OBSERVER Department of Electrical Engineering Southern Taiwan University (a) structure of WFNN (b) triangular functions in membership layer of WFNN; (c) triangular functions in wavelet layer of WFNN Department of Electrical Engineering Southern Taiwan University Description of WFNN The j th fuzzy rule of the proposed WFNN can be presented as follows 1 1 j 1 1 2 R j : if x is M and x is M j 2 then j is w ikik xi1 , i 1,2 Where i R j jth rule of WFNN j internal variable xi1 input of WFNN M 1j , M 2j input of WFNN ik ith in kth term wavelet output to node of wavelet sum layer wik Wavelet we ights for units in wavelet mechanism layer Department of Electrical Engineering Southern Taiwan University The signal propagation and the basic function in each layer of the WFNN are : Layer 1: Input Layer For every node in this layer y f x N x N , i 1,2 1 i Where 1 1 i 1 i y output of membership layer 1 1 i x e Eˆ sin re ˆre 1 1 x e 1 2 Department of Electrical Engineering Southern Taiwan University (the angle estimation error term of the rotor flux angle) (the derivative of the angle estimation error term) Layer 2: Membership Layer y 2j N net 2j N j 1,2 6 1 1 0 if y m , y i j j i mj j y i1 m j j if m j j y i1 , y i1 m j j y i1 m j j if m j y i1 m j j j Where y output of membership layer 2 2 j Department of Electrical Engineering Southern Taiwan University Layer 3: Rule Layer and Wavelet Mechanism y k3 N net k3 N w 3jk y 2j N , k 1, ,9 j Where the node in the layer j w connecting weight between 3 jk membership layer and rule layer Department of Electrical Engineering Southern Taiwan University 0 1 x m j 4 j 1 i j 4 j 1 1 xi m j 2 j 2 j ik xi1 j xi1 m j 2 j 1 j 2 j 1 1 xi m j 4 j 4 j j if xi1 m j 4 j , xi1 m j 4 j Where if m j 4 j xi1 m j 2 j if m j 2 j xi1 m j if m j xi1 m j 2 j if m j 2 j xi1 m j 4 j k wikik x k kth term wavelet mechanism output to node of wavelet layer Department of Electrical Engineering Southern Taiwan University Layer 4: Consequent Layer y N net N k w y N , l 1, ,9 4 l 4 l 4 k 3 k Layer 5: Output Layer y N net N w y N , o 1 5 o 5 o 5 l i Department of Electrical Engineering Southern Taiwan University 4 l Online Learning Algorithm for WFNN The online parameter learning is based on the supervised learning algorithms to adjust the link weights in layer 5, the link weights in layer 3, and the parameters of membership functions in layer 2 using the back-propagation (BP) algorithm to minimize the following energy function: 1 2 V e 2 Department of Electrical Engineering Southern Taiwan University Layer 5 The error term V 5 yo 5 o Update rule 5 V y V wl5 w1 5 w1 5 o5 w1 o5 yl4 wl yo wl Department of Electrical Engineering Southern Taiwan University Layer 4 The error term Vy V 5 5 4 5 o wl yl yo y 4 l Department of Electrical Engineering Southern Taiwan University 5 o 4 l Layer 3 The error term 5 4 V y y V k3 3 5 o4 l 3 l4 wk4 k y k y o yl y k Update rule wik w 2 V w2 wik Vy o5 yl4 y o5 yl4 wik w2 l4 y k3 wk41k , i 1 4 3 4 w2 l y k wk 2 k , i 2 Department of Electrical Engineering Southern Taiwan University Layer 2 The error term Vy y y V 3 3 2 5 k yk y j y o y y y k 5 o 4 l 2 j Update rule of m j 4 l 3 k 3 k 2 j Vyo5 yl4yk3y 2j V m j m m 5 4 3 2 m j yo yl yk y j m j 1 1 0 if y m , y i j j i mj j 2 1 m j if m j j y i1 m j j 2 1 1 if m y j i mj j m j j Department of Electrical Engineering Southern Taiwan University Update rule of j Vy o5 y l4 y k3 y 2j V j 5 4 3 2 j y o y l y k y j j 0 1 y i mj 2 j 2 j 1 y i mj 2 j 2j Department of Electrical Engineering Southern Taiwan University if y i1 m j j , y i1 m j j if m j j y i1 m j if m j y i1 m j j Where Department of Electrical Engineering Southern Taiwan University The weights, mean, and standard deviation of the membership functions are updated as follows: w N 1 w N w 5 l 5 l 5 l wik N 1 wik N wik m j N 1 m j N m j j N 1 j N m j Department of Electrical Engineering Southern Taiwan University DESIGN AND EXPERIMENTATION Photos of experimental drive system and plant. (a) IPMSM drive system. (b) Motor test platform Department of Electrical Engineering Southern Taiwan University The IPMSM manufactured by Rechi Precision Co., Ltd., Taiwan The IPMSM is a three-phase Y-connected 4-pole 945 W 220V/6.87A 4000 rpm 22 kg-cm type. The parameters of the motor at the nominal condition are given as follows: r s 0.34 , L d 4.43 mH L q 9.26 mH , J 0.0005 Nxm/ rad/s 2 , B 0.009 Nxmrad/s , pm 0.11.3 Wb Department of Electrical Engineering Southern Taiwan University The gains of the PI speed controller, the gains of the PI current controller, and the gains of PID torque observer are given as follows: K wp 0.0348 , K wi 0.5512 , K rp 2.3166, K ri 398.288 , K p 0.0464 , K i 0.3662, K d 0.0006 Department of Electrical Engineering Southern Taiwan University The rotor flux angle estimation error and the speed error of speed command tracking are defined as re _ error N ˆre N re ( N ) rm _ error N Department of Electrical Engineering Southern Taiwan University * rm N rm ( N ) Simulated Results of Saliency Back-EMF-Based MTPA Control (a) Speed responses and speed estimation error of saliency back-EMF-based rotor flux angle and speed estimator. (b) Current responses of sensorless control without MTPA. (c) Current responses of conventional MTPA sensorless control. (d) Current responses of proposed saliency back-EMF-based MTPA sensorless control. Department of Electrical Engineering Southern Taiwan University Experimental Results of Saliency BackEMF-Based MTPA Control (a) Speed and current responses from 3500 to 4000 rpm without MTPA control at Case 1. (b) Speed and current responses from 3500 to 4000 rpm without MTPA control at Case 2. (c) Speed and current responses from 3500 to 4000 rpm with MTPA control at Case 1. (d) Speed and current responses from 3500 to 4000 rpm with MTPA control at Case 2. Department of Electrical Engineering Southern Taiwan University Experimental results of saliency back-EMFbased MTPA sensorless control (a) Speed and current responses from 0 to 4000 rpm with PID torque observer at Case 1. (b) Speed and current responses from 0 to 4000 rpm with PID torque observer at Case 2. (c) Speed error, input, and output of PID torque observer at Case 1. (d) Speed error, input, and output of PID torque observer at Case 2. Department of Electrical Engineering Southern Taiwan University Experimental results of saliency back-EMF-based MTPA sensorless control (e) Speed and current responses from 0 to 4000 rpm with FNN torque observer at Case 1. (f) Speed and current responses from 0 to 4000 rpm with FNN torque observer at Case 2. (g) Speed error, input, and output of FNN torque observer at Case 1. (h) Speed error, input, and output of FNN torque observer at Case 2. Department of Electrical Engineering Southern Taiwan University Experimental results of saliency back-EMF-based MTPA sensorless control using WFNN torque observer (a) Speed and current responses from 0 to 4000 rpm at Case 1. (b) Speed and current responses from 0 to 4000 rpm at Case 2. (c) Speed error, input, and output of WFNN torque observer at Case 1. (d) Speed error, input, and output of WFNN torque observer at Case 2. Department of Electrical Engineering Southern Taiwan University Comparison of PID, FNN and WFNN torque observers of saliency back-EMF-based MTPA sensorless control at Case 2 (a)–speed and rotor flux angle estimation error using PID torque observer from 3000 to 3500 rpm, (b)–speed and rotor flux angle estimation error using FNN torque observer from 3000 to 3500 rpm, (c)–speed and rotor flux angle estimation error using WFNN torque observer from 3000 to 3500 rpm Department of Electrical Engineering Southern Taiwan University Comparison of PID, FNN and WFNN torque observers of saliency back-EMF-based MTPA sensorless control at Case 2 (d)–performance measurings of rotor flux angle estimation error, and (e)–performance measurings of speed estimation error Department of Electrical Engineering Southern Taiwan University CONCLUSION The WFNN torque observer has been proposed to replace the traditional PID observer used in the conventional saliency back-EMF-based rotor flux angle and speed estimator to improve the estimating performance of the rotor flux angle and speed The simulated and experimental results have verified the proposed saliency back-EMF-based rotor flux angle and speed estimator using WFNN torque observer combine with a new MTPA control resulted in better estimating and control performance than the conventional saliency back-EMF-based rotor flux angle and speed estimator. Department of Electrical Engineering Southern Taiwan University REFERENCES C. T. Pan and S. M. Sue, “A linear maximum torque per ampere control for IPMSM drives over full-speed range”, IEEE Trans. Energy Convers., vol. 20, no. 2, pp. 359366, Jun., 2005, IEEE. 2. A. Consoli, S. Musumeci, A. Raciti and A. Testa, “Sensorless vector and speed control of brushless motor drives”, IEEE Trans. Ind. Electron., vol. 41, no. 1, pp. 91-96, Feb., 1994, IEEE. 3. C. French and P. Acarnley, “Control of permanent magnet motor drives using a new position estimation technique”, IEEE Trans. Ind. Appl., vol. 32, no. 5, pp. 10891097, Sep./Oct., 1996, IEEE 1. Department of Electrical Engineering Southern Taiwan University